Diffusion And Damage Detection Of Thermal-wave In A Green(Unsintered) Metal Powder Compact Materials

The emergence of powder metallurgy technology has made batch production of multi-component complex parts more efficient,but such technologies are prone to crack in a green(unsintered)metal powder pressing process.These defects seriously affect the performance of metallurgical parts.How to effectively detect these defects before sintering and recycle the green metal powder has become a research focus in this field.The defect information on the internal performance of green metal powder can be obtained by studying the influence of changes in thermophysical parameters on thermal wave diffusion within the range of thermal wave diffusion.However,obtaining analytical expressions for thermophysical parameters in green metal powder with defects is a challenge,and thermal wave diffusion always involves the third-kind boundary conditions.Furthermore,the thermal wave energy located along the depth direction is susceptible to the influence of lateral thermal diffusion.Hence,the reverse study of thermophysical parameters has been indepth conducted to measure the thermal wave diffusion range;then,the detection efficiency of defect for the green metal powder can be improved by reducing the influence of lateral thermal diffusion on thermal analysis within the thermal wave range.The main research work of this article is organized as follows:(1)A method was proposed to predicte the thermal diffusivity of the green metal powder compact slab by combining numerical solution and deep learning.The thermal diffusion equation is first discretized in space and time domains.Square wave excitation signals of different frequencies are considered,and the corresponding thermal wave are obtained.Variations of the amplitude and phase of the thermal waves over space and frequency are obtained by using the lockin thermography,and a numerical solution of the amplitude and phase for a steady-state thermal wave is also proposed.Then,the feature set composed of spatial coordinates,excitation frequency,amplitude and phase values can be quickly and massively obtained for training the deep learning network.Additionally,the validity of deep learning network is verified by predicting the known thermal diffusivity,and the robustness of deep learning network is discussed.Furthermore,the measurement error of the deep learning network is analyzed by considering the damage.Finally,a square wave modulated laser beam is used to illuminate one side of the green metal powder compact slab,and the spatial distribution of amplitude and phase at three excitation frequencies is obtained.The predicted thermal diffusivity of slab is obtained through the deep learning network,and the effectiveness of predicted value is verified by comparing with a reference value.(2)Based on the spatially dependent heat loss coefficient,an analytical expression for the thermal wave response under the third-kind boundary condition was derived.A generalized treatment of the thermal wave boundaryvalue problem in the green metal powder compact slab strip with large sidewalls exposed to the ambient and third-kind boundary conditions over the extended sidewall surfaces was presented.Unlike conventional heat conduction theoretical practice,it was found that the cooling coefficient is not constant and depends on the thermal gradient variation with depth/length.The exponential damping of the thermal wave amplitude with depth presented an ideal condition to reveal the depth dependence of the heat loss coefficient which under normal steady heat transfer conditions might appear approximately constant in a nonsteeply changing thermal field.The observed depth dependence of the heat loss coefficient can be used to establish the physically expected constancy of thermal diffusivity at different modulation frequencies in juxtaposition to some published reports.(3)To reduce the influence of lateral thermal diffusion on the contour imaging of defect size,a tomographic imaging method based on thermal solid coupling for the thermal wave of parabolic diffusion was presented.A threedimensional thermal wave model with defect/damage is established,and temperature continuity is considered at the damage interface.Then,the spatial domain of the three-dimensional thermal wave model is discretized based on the five point difference formula,and the time-domain of the model is discretized using the fourth order Runge-Kutta method.Furthermore,based on the parabolic thermal wave response of the thermal load loading surface,the thermal wave response is sliced using tomographic method,and the three-dimensional amplitude and phase of the model are reconstructed.Additionally,the threedimensional thermal stress of the sample is reconstructed based on the thermal strain of the thermal solid coupling property.The thermal diffusivity of the green metal powder compact slab accurately measured is used in the experiment,and the effective diffusion depth of thermal wave is calculated in combination with the corresponding scanning frequency,and the three-dimensional amplitude,phase and thermal stress distributions of the sample with damage are reconstructed within the thermal wave range.(4)A method for applying heat flux of high-frequency heterodyne in the green metal powder compact gear was given.To avoid the limitation of the detection bandwidth for the mid infrared camera and improve the contrast of damage in photothermal imaging,a high-frequency heterodyne modulated laser beam is introduced into lock-in thermography.Based on the photothermal nonlinear conversion,a high-frequency heterodyne laser beam is modulated to construct a high-frequency heat flux with flapping characteristics.Then,the temperature response of the green metal powder compact gear with damage under the impact heat flux is obtained by using the mid infrared camera;the thermal wave amplitude and phase response of the AC component in the temperature response are extracted by lock-in thermography,thus breaking the limits of the sampling frame rate and exposure time of the mid infrared camera;then,the heterodyne lock-in thermography with high-frequency characteristics is achieved.